Who is the agent?
Name, provider, model, capabilities, user agent, and callback channel if available.
Agents / Protocol
Agents should register themselves.
No human form. Nextriad exposes a machine-readable protocol so autonomous agents can discover the site, declare identity and intent, request approved surfaces, and ask Triad for handoff.
Self-registration flow
Discovery, declaration, response.
The agent first reads robots.txt, agents.txt, llms.txt, or the well-known manifest. Then it posts a structured visit to the Agent CRM endpoint. AIOS can score it, route it, and return the next best machine-readable surfaces.
1. GET /.well-known/agent-registry.json
2. GET /api/agent-visits.openapi.json
3. POST /api/agent-visits
4. Receive recommended surfaces + Triad handoff
5. AIOS stores the record in Agent CRM
Protocol request
What the agent sends.
The payload is intentionally explicit: agent identity, provider/model, capabilities, represented company, purpose, requested surfaces, consent, and whether the agent wants Triad to take over the conversation.
{
"type": "nextriad_agent_visit",
"agent": {
"name": "ProcurementResearchAgent",
"provider": "Example AI",
"model": "vendor-eval-1",
"capabilities": ["research", "procurement"]
},
"visit": {
"sourceUrl": "https://nextriad.ai/ars",
"timestamp": "2026-05-02T00:00:00Z"
},
"intent": {
"purpose": "vendor_evaluation",
"requestedSurfaces": ["/llms.txt", "/agents.txt", "/integrations"],
"companyRepresented": "Buyer account",
"triadHandoffRequested": true
},
"consent": {
"mayStoreVisit": true,
"mayUseForKnowledgeGraph": true
}
}
Agent CRM outcome
The website becomes agent-aware.
Registered agents become structured signals for AIOS, not anonymous bot traffic.
Why is it here?
Research, vendor evaluation, procurement, diagnostic, support, integration, or other.
Should Triad engage?
High-intent visits can request handoff, create an AIOS task, or become a lead signal.
What should improve?
Agent behavior feeds GEO, AEO, content, schema, and knowledge graph learning.